Impact of allogeneic hematopoietic stem cell transplantation on pediatric acute myeloid leukemia of FAB subtypes M4 and M5

Background: FAB-M4 and M5 are unique subgroups of pediatric acute myeloid leukemia. However, for these patients, few studies have demonstrated the clinical and biological characteristics and ecacy of hematopoietic stem cell transplantation (HSCT), and especially haplo-HSCT. Procedure: We retrospectively evaluated the outcomes of 70 children with FAB-M4/M5 enrolled in our center from January 2013 to December 2017. Results: Of the patients, 32, 23, and 15 were in low-risk, intermediate-risk, and high-risk groups, respectively. T(16;16), inv16/CBFB-MYH11 was the most frequent cytogenetic abnormality. Among detected genetic alterations, WT1 was mutated at the highest frequency, followed by FLT3-ITD, NPM1, and CEBPA. Thirty-three patients received HSCT (haplo-HSCT = 30), of which four, 18, and 11 were in low-risk, intermediate-risk, and high-risk groups, respectively. For all patients, the 3-year overall survival (OS), event-free survival (EFS), and cumulative incidence of relapse (CIR) were 85.3 ± 4.3%, 69.0 ± 5.7%, and 27.9 ± 5.2%, respectively. By multivariate analysis, low-risk stratication predicted superior OS, EFS, and PLT ≤ 50 × 10 9 /L at diagnosis, with FLT3-ITD mutations predicting higher CIR and poorer EFS. In intermediate- and high-risk groups, HSCT was independently associated with higher EFS and lower CIR. With a median post-transplant observation time of 30.0 months, the 3-year OS, EFS, CIR, and non-relapse mortality in the haplo-HSCT group were 74.2 ± 8.6%, 68.3 ± 8.9, 24.6 ± 7.6%, and 6.6 ± 4.1%, respectively. Conclusions: Risk-oriented treatment is important for pediatric FAB-M4/M5. For intermediate- and high-risk groups, HSCT signicantly improved survival and haplo-HSCT

shows the overall pro le of the enrolled patients, including risk classi cation and donor availability.

Treatment protocols
All patients received 1-2 cycles of induction chemotherapy of I(D)AE, including idarubicin/daunorubicin, cytarabine, and etoposide. Consolidation chemotherapy regimens included HD Ara-c, HA, and I(D)AE. During the consolidation period, these three regimens were consecutively applied for 12-18 months, and then withdrawal follow-up began. A total of 4-6 rounds of HD Ara-c were administrated during the entire consolidation treatment and the cumulative dose of anthracyclines was equivalent to DNR 350 mg/m 2 . All patients received intrathecal chemotherapy with methotrexate (MTX), cytarabine, and dexamethasone regularly as central nervous system leukemia (CNSL) prophylaxis. Over the total treatment course, intrathecal treatment was performed 4-8 times. The detailed chemotherapy protocol is shown in (Supp. We used granulocyte colony-stimulating factor-mobilized bone marrow cells plus peripheral blood stem cells as a graft resource. In matched sibling transplants, the pre-conditioning treatment was a modi ed BU/CY regimen (busulfan-cyclophosphamide), which included the following: hydroxyurea (80 mg/kg per day, orally) on day −10; Ara-c (2 g/m 2 per day, intravenously) on day −9; busulfan (Bu, 3.2 mg/kg per day, intravenously) on days −8 to −6; cyclophosphamide (Cy, 1.8 g/m 2 per day, intravenously) on days −5 to −4; methyl-N-(2-chloroethyl)-N-cyclohexyl-N-nitrosourea (Me-CCNU, 250 mg/kg per day, orally) on day −3. For haplo-transplants, the pre-conditioning treatment was a modi ed BU/CY regimen combined with antihuman thymocyte immunoglobulin (ATG), consisting of the following: Ara-c (4 g/m 2 per day, intravenously) on days −10 to −9; Bu, Cy, and Me-CCNU (same as previous described); ATG (2.5 mg/kg per day, intravenously) on days −5 to −2. The stem cells for haplo-transplant were derived from unmanipulated bone marrow and not involved in the CD34 selection process. All patients in the transplant group received cyclosporin A, mycophenolatemofetil, and short-term MTX as acute graft-versus-host disease (aGVHD) prophylaxis. The details about pre-conditioning regimens and supportive care were described previously [15].
De nitions and assessments CR was de ned as bone marrow blasts < 5% with complete hematologic recovery (neutrophil count > 1.0 × 10 9 /L, platelet count > 80 ×10 9 /L, independence of red cell transfusions) and absence of extramedullary disease; Partial remission was de ned as bone marrow blasts between 5% and 20% after the rst induction cycle; non remission was de ned as bone marrow blasts > 20% after the rst cycle of induction. Relapse was de ned as bone marrow blasts ≥ 5%, the reappearance of blasts in the blood, or development of extramedullary disease. Minimal residual disease (MRD) was assessed by multiparameter ow cytometry (MFC) [16] and PCR-based evaluation of the expression levels of speci c fusion genes. MRD-positive was de ned as two consecutive positive results by MFC or a speci c fusion gene, or both MFC and fusion genes positive in a single sample [15]. Patients with MRD from negative to positive were not classi ed as having relapsed. Overall survival (OS) was calculated from the date of diagnosis to the date of death from any cause or the date of last contact. EFS was calculated from the date of diagnosis to the date of rst event (relapse, second malignancy, or death due to any cause, whichever occurred rst) or the date of last follow-up. When analyzing the OS, EFS, and relapse rate in the haplo-HSCT group, the initial time was calculated from the date of transplantation.

Statistical analysis
The last follow-up date was January 01, 2020. Probabilities of OS and EFS were estimated using the Kaplan-Meier method. The cumulative incidence of GVHD, relapse, and non-relapse mortality (NRM) were calculated using competing risk analyses. We used Cox proportional hazards regression to estimate the multivariate hazard ratios (HRs) for OS and EFS and used competing risk regression analyses to estimate the multivariate HRs for relapse. Variables with P values < 0.1 in univariate analysis were included in multivariate analysis. A two-sided P value of 0.05 was considered signi cant. Data analyses were performed using the SPSS software package (SPSS Inc., Chicago, IL, USA) and SAS (SAS Institute, Cary, NC).

Clinical outcomes of all patients
Of the 70 patients enrolled, 48 (68.6%) achieved CR after the rst cycle of induction therapy, 21 (30.0%) achieved CR after the second cycle of induction therapy, and one (1.4%) achieved CR after three courses of chemotherapy. Relapse occurred in 19 (24.1%) patients, including 18 hematologic relapses and one hematologic combined with extramedullary leukemia relapse. The median time from diagnosis to relapse was 10.3 months (range 3.1-35.2). Ten of the 19 relapsed patients died of disease progression, whereas one is alive with disease and eight are alive without disease in the second CR. Up to the last follow-up time, the median follow-up time was 37.1 months (range 4.9-84.0). Twelve patients had died (10 due to relapse and two due to transplant-related mortality). For all 70 patients, the cumulative incidence of relapse (CIR) at 3 years was 27.9 ± 5.2%, the cumulative incidence of NRM at 3 years was 3.2 ± 2.2%, and the probability rates of OS and EFS at 3 years were 85.3 ± 4.3% and 69.0 ± 5.7%, respectively.

Overall survival
In univariate analysis (Supp. Fig. 1, Supp.     , 102-647). The cumulative incidences of cGVHD and extensive cGVHD at 3 years were 56.8 ± 7.6% and 16.4 ± 3.2%, respectively. Nineteen (63.3%) patients developed cytomegaloviremia, and the cumulative incidences of cytomegaloviremia at day 100 after HSCT was 64.9 ± 5.7%. Seven (23.3%) patients experienced hemorrhage cystitis, and the cumulative incidences of total hemorrhage cystitis and grade III-IV hemorrhage cystitis at day 100 after HSCT were 28.7 ± 6.2% and 6.7 ± 1.4%, respectively. During the follow-up period after HSCT, seven patients experienced relapse with a median time from HSCT to relapse of 243 days (range, 80-824). The CIR at 3 years was 24.6 ± 7.6% (95% CI: 13.4-45.0%). Two patients died due to non-relapse factors after transplantation, one died of multiple organ failure 235 days post-transplantation, and one died of intracranial infection 8 days post-transplantation. The NRM at 3 years was 6.6 ± 4.1% (95% CI: 1.9-22.6%). At the time of the last followup, seven patients ( ve due to relapse and two due to transplant-related mortality) had died with a median time to death of 235 days (range, . The probability rates of OS and EFS at 3 years after HSCT for all 30 children were 74.2 ± 8.6% and 68.3 ± 8.9%, respectively (Supp. Fig. 2).

Discussion
Whether it was research on clinical e cacy or molecular biology, previous studies have mainly focused on overall pediatric AML. However, research on pediatric FAB-M4/M5, a clinically and biologically heterogeneous disease, is hardly reported. Here, we provide the clinical characteristics and molecular genetics of pediatric FAB-M4/M5 and explored the e cacy of HSCT, especially haplo-HSCT, for these patients. FAB-M4/M5 comprises 31.5% of total childhood AML in our center. Nearly half of the patients were older than 10 years and had a WBC count more than 50 × 10 9 /L at diagnosis. Further, 14.3% of patients had CNSL. These clinical features are consistent with previous reports [3,8,17]. In this retrospective study, the 3-year OS and EFS of children with FAB-M4/M5 were 85.3% and 69.0%, respectively, which were similar to results reported by Imamura et al [18] and also similar to the results of overall pediatric AML reported in other centers [4,[18][19][20].
The signi cant improvements in the prognosis of pediatric AML are mainly attributed to re nements in cytogenetics-based risk strati cation and subsequent risk-directed therapy. Therefore, comprehensive molecular pro ling at diagnosis is essential. Existing evidence has shown the precise prognostic signi cance of some certain gene fusions and mutations [10,[20][21][22][23]. Including CBF, NPM1, and CEBPA mutations, the consensus low-risk group currently accounts for approximately 30% to 40% of overall pediatric AML and is associated with a favorable prognosis [9,10,14,24]. Due to the more frequent occurrence of t(16;16), inv 16/CBFB-MYH11 in pediatric FAB-M4/M5 [25], the proportion of FAB-M4/M5 patients in the low-risk group might be much higher than that of overall childhood AML. The low-risk group comprised 47.2% of all children with FAB-M4/M5 in this study and had an excellent prognosis with 3-year OS and EFS probabilities of 96.9% and 87.3%, respectively. As a common marker of the high-risk group, FLT3/ITD mutations occur in 10% to 20% of children with AML [14] and often confer poor survival [12,26]. We found a similar result in our study cohort. Patients with FLT3/ITD mutations accounted for 10.0% of overall FAB-M4/M5 cases and these were identi ed as an independent risk factor for relapse and survival. We also provide several possible causes of deviations in the results and limitations. First, the frequency of HSCT during CR1 in our study was similar to that in the study of Horan et al [27] but was much higher than that at other centers, which might lead to bias [18,30]. Second, although HSCT showed obvious advantages for the high-risk group, the survival rate of these patients was not good. We consider that the small number of cases in the high-risk group available for this analysis limited the ability to draw any de nitive conclusions, but our results do suggest that even with HSCT, these patients have poor   Figure 1 Overall pro le of the enrolled patients including risk classi cation and donor availability Figure 2